The future of modern day technology and particularly computer technology demands smaller, faster and more reliable electronic devices. Unfortunately, simply extending existing technology by forming devices having smaller dimensional features presents limitations. Once feature sizes of electronic devices pass below 100 nm (nanometers), a fundamental shift in operation can be expected due to quantum mechanical effects. At these nanoscale dimensions, quantum interference effects will dominate. However, conventional device technology has not yet attained nanoscale features in which quantum effects dominate.
Yet quantum interference devices (QIDs) continue to receive serious attention because of their potential for unique transistor devices with excellent power delay products. Some QID structures could be turned on optically to circumvent the RC time constant and carrier lifetime limitations. For such devices, switching times of one picosecond and power delay products of 200 fJ (femto Joules) are predicted.
Quantum well structures have been widely demonstrated in thin planar structures. Quantum wires and dots require the fabrication of nanoscale features onto these planar structures. To date, researchers have fabricated features as small as 100 nm by lithographic techniques. However, because of the large amount of time required to form a feature, such lithographic techniques are generally unsuitable for large scale fabrication of nanoscale features.
One significant problem related to the fabrication of nanoscale features in surfaces of materials is process induced damage to the underlying region of the material. For example, process induced damage in semiconductors is a source of great concern. Recent studies indicate that what has been viewed as subsurface damage can be attributed to secondary effects such as inhomogeneous surface contamination. Other studies have shown that an inhomogeneous oxide can lead to irregular etching of a semiconductor surface which had been previously attributed to subsurface damage.
Mechanical material removal processes generally can be reduced to an energetics argument. In "Ductile-Regime Grinding of Brittle Materials," by T. G. Bifano, SPIE, Vol. 966 (1988), it was found that material removal from a surface of a material fundamentally changes as the volumetric removal rate is decreased. The energy required for brittle fracture (E.sub.f) of the material is: EQU E.sub.f =(G) (A.sub.f)
where,
G= the Griffith crack propagation parameter and PA1 A.sub.f = the area of the new surface created. PA1 .sigma..sub.y = the yield stress and
The energy required for plastic deformation (E.sub.p) of the material is: EQU Ep= (.sigma..sub.y) (V.sub.p)
where,
Vp = the volume of material to be plastically deformed. Since V.sub.p is proportional to d.sup.3 and A.sub.f is proportional to d.sup.2, where d is a distance related to removal depth, then E.sub.p is proportional to d.sup.3 and E.sub.f is proportional to d.sup.2. Therefore, as the depth (d) of the material removal process is reduced, the energy required for brittle fracture (E.sub.f) becomes greater than the energy required for ductile material removal (E.sub.p). Once this condition occurs, the material removal shifts from brittle to predominately ductile material removal. This behavior has been confirmed in grinding studies on GaAs and InP (see Journal of Elec. Chem. Soc. Vol. 138, p. 1826, (1991)).
By convention, researchers have assumed that the depth of subsurface damage resulting from material removal processes such as lapping, grinding or polishing is roughly equal to the diameter of the grit. However, a study on silicon conducted at Uppsala University has shown an exponential decrease in subsurface damage as grit size is reduced (see "Micromechanical Properties of Silicon," by S. Johansson, 1988). In the study, it was shown that when the grit size was reduced to 3 microns, the depth of subsurface damage was confined to 50 nm. At grit sizes of 50-70 nm no damage could be measured by cross-sectional transmission microscopy (TEM). In a study of erosion rates on silicon and gallium arsenide surfaces, these same researchers claimed to observe an elastic material removal process. It is to this general problem that the present invention is directed.